CS 415: Programming Languages Chapter 1 Aaron Bloomfield Fall 2005 The first computers Scales – computed relative weight of two items Computed if the first item’s weight was less than, equal to, or greater than the second item’s weight Abacus – performed mathematical computations Primarily thought of as Chinese, but also Japanese, Mayan, Russian, and Roman versions Can do square roots and cube roots Stonehenge Computer Size ENIAC then… ENIAC today… With computers (small) size does matter! Why study programming languages? Become a better software engineer Understand how to use language features Appreciate implementation issues Better background for language selection Familiar with range of languages Understand issues / advantages / disadvantages Better able to learn languages You might need to know a lot Why study programming languages? Better understanding of implementation issues How is “this feature” implemented? Why does “this part” run so slowly? Better able to design languages Those who ignore history are bound to repeat it… Why are there so many programming languages? There are thousands! Evolution Structured languages -> OO programming Special purposes Lisp for symbols; Snobol for strings; C for systems; Prolog for relationships Personal preference Programmers have their own personal tastes Expressive power Some features allow you to express your ideas better Why are there so many programming languages? Easy to use Especially for teaching / learning tasks Ease of implementation Easy to write a compiler / interpreter for Good compilers Fortran in the 50’s and 60’s Economics, patronage Cobol and Ada, for example Programming domains Scientific applications Using the computer as a large calculator Fortran and friends, some Algol, APL Using the computer for symbol manipulation Mathematica Business applications Data processing and business procedures Cobol, some PL/1, RPG, spreadsheets Systems programming Building operating systems and utilities C, PL/S, ESPOL, Bliss, some Algol and derivitaves Programming domains Parallel programming Parallel and distributed systems Ada, CSP, Modula, DP, Mentat/Legion Artificial intelligence Uses symbolic rather than numeric computations Lists as main data structure Flexibility (code = data) Lisp in 1959, Prolog in the 1970s Scripting languages A list of commands to be executed UNIX shell programming, awk, tcl, Perl Programming domains Education Languages designed to facilitate teaching Pascal, BASIC, Logo Special purpose Other than the above… Simulation Specialized equipment control String processing Visual languages Programming paradigms You have already seen assembly language We will study five language paradigms: Top-down (Algol 60 and Fortran) Functional (Scheme and/or OCaml) Logic (Prolog) Object oriented (Smalltalk) Aspect oriented (AspectJ) Programming language history Pseudocodes (195X) – Many Fortran (195X) – IBM, Backus Lisp (196x) – McCarthy Algol (1958) – Committee (led to Pascal, Ada) Cobol (196X) – Hopper Functional programming – FP, Scheme, Haskell, ML Logic programming – Prolog Object oriented programming – Smalltalk, C++, Python, Java Aspect oriented programming – AspectJ, AspectC++ Parallel / non-deterministic programming Compilation vs. Translation Translation: does a ‘mechanical’ translation of the source code No deep analysis of the syntax/semantics of the code Compilation: does a translation of the code thorough understanding and A compiler/translator changes a program from one language into another C compiler: from C into assembly An assembler then translates it into machine language Java compiler: from Java code to Java bytecode The Java interpreter then runs the bytecode Compilation stages Scanner Parser Semantic analysis Intermediate code generation Machine-independent code improvement (optional) Target code generation Machine-specific code improvement (optional) For many compilers, the result is assembly Which then has to be run through an assembler These stages are machine-independent! The generate “intermediate code” Compilation: Scanner Recognizes the ‘tokens’ of a program Example tokens: ( 75 main int { return ; foo Lexical errors are detected here More on this in a future lecture Compilation: Parser Puts the tokens together into a pattern void main ( int argc , char ** argv ) { This line has 11 tokens It is the beginning of a method Syntatic errors are detected here When the tokens are not in the correct order: int int foo ; This line has 4 tokens After the type (int), the parser expects a variable name Not another type Compilation: Semantic analysis Checks for semantic correctness A semantic error: foo = 5; int foo; In C (and most languages), a variable has to be declared before it is used Note that this is syntactically correct As both lines are valid lines as far as the parser is concerned Compilation: Intermediate code generation (and improvement) Almost all compilers generate intermediate code This allows part of the compiler to be machineindependent That code can then be optimized Optimize for speed, memory usage, or program footprint Compilation: Target code generation (and improvement) The intermediate code is then translated into the target code For most compilers, the target code is assembly For Java, the target code is Java bytecode That code can then be further optimized Optimize for speed, memory usage, or program footprint